We are 14 per cent more likely to die on our birthday, according to new research based on a study of more than two million people.

The authors of the study blame what they have called anniversary reaction hypothesis, or birthday blues, with stress playing a significant part.

Causes of birthday deaths included heart attacks, stroke, cancer, falls and suicide.

“We concluded that birthdays end lethally more frequently than might be expected,’’ the Independent quoted Dr Vladeta Ajdacic-Gross, who led the Swiss study, as saying.

The researchers analysed 2.4 million deaths over a 40-year period. Results show there were 13.8 per cent more deaths on birthdays than might be expected compared with any other day of the year.

The risk increased with age, with the figure rising to 18 per cent among the over 60s. Results for individual diseases show that there was an 18.6 per cent increased risk of dying from cardiovascular disease, including heart attacks, and a higher risk of up to 21.5 per cent for strokes.

There was also a 10.8 per cent rise in deaths among people with cancer.

Among men, there was a 34.9 per cent rise in suicides, and a 28.5 per cent increase in accidental deaths.

There was also a 44 per cent rise in fatal falls on birthdays, with the risk increasing from about four days before a birthday.

The researchers said that there are two competing explanations for the phenomenon.

The postponement theory suggests that seriously ill people can somehow hold on until their birthday has passed, as they try to reach the milestone.

However, the researchers said that this theory is not supported by their findings. They support the anniversary hypothesis, which maintains it is the event itself that is responsible.

In the case of heart attacks and strokes, for instance, stress may be to blame.

Older people, in particular, they suggest, may feel exposed to stress on birthdays. The research indicates that hospital admissions for cardiovascular conditions occur more frequently on birthdays.

The study was published in Annals of Epidemiology.